Recently, in accordance with developments of the Internet, systems with which contents such as knowledge and opinions of individuals are shared by many people have been operated by using websites on the Internet. Examples thereof are knowledge bases which manage common knowledge regarding services and operations in business firms and word-of-mouth sites operated by accumulating word-of-mouth information of consumers.
As a means for viewers to search contents in the above-described systems, it is common to designate attributes given to the contents in advance and keywords contained in the contents and to execute full-text search. Further, when contents creators register the contents, the creators have been trying to support viewers to be able to collect information efficiently by putting references to the contents that are related to the contents created by the creators themselves. However, there are followings issues with such systems.
First, since the contents are created by individuals, those contents are created based on the creator's personal viewpoint. Therefore, there is a possibility that the attributes added by the creator to the contents and words and the like used in expressions of the contents do not match with those conjectured by a system manager and a viewer. In such case, for example, if the viewer executes category designation, keyword designation, full-text search, etc., for finding target contents, the viewer cannot find the contents even if there are the contents desired by the viewer existing in the system.
Further, when the relevancy between contents is designated manually, relevancy between the contents can only be given at limited timing such as when the contents creator register or correct the contents. In that case, setting of the relevancy between already-existing contents and newly registered contents may be missed out.
For such issue, the system disclosed in Patent Document 1 monitors access to information made by a user is monitored at all times to create an access history database, and extracts information having an access history near the time at which a reference document is accessed from the database. Then, the extracted information is outputted to a screen as the information related to the reference document. Furthermore, Patent Document 1 also discloses a method for evaluating the relevancy according to access time intervals with respect to the reference document.    Patent Document 1: Japanese Unexamined Patent Publication H10-11458
However, with the technique described above, there may be cases where the relevancy between the contents is not necessarily adequate. For example, when a user conducts a plurality of searches simultaneously by utilizing websites on the Internet, the relevancy between information pieces may be low even those are the information pieces accessed at short-time interval from each other. If so, information having adequate relevancy with respect to specific information cannot be extracted, so that it is not possible to search information and provide information efficiently.